Data from the Linked Birth/Infant Death files prepared by the National Center for Health Statistics are being used to calculate race-specific relative risk and population attributable risks for five adverse outcomes (low birth weight, very low birth weight, neonatal death, post-neonatal death, and infant death) in twin pregnancy. The role of the former two intermediate outcomes in explaining the later three outcomes will also be assessed. Because the data set used is population-based for the entire nation, the risk measures reported will be free from referral biases that frequently distort risk estimates that originate from teaching or referral hospitals.